Pete Willemsen (pwillems++at++ford.com)
Tue, 8 Jul 1997 09:31:56 -0400
In regards to double buffering on the IR, where does the memory for
the "back" buffer come from? I've been doing some calculations
regarding whether or not some of our desired configurations would be
possible on our 1 RM-6. In doing this I became confused as to whether
the memory for the "back" buffer should be included in the
calculation.
My confusion also lead me to wonder if Performer is automatically
using double buffering. I do not recall specifying that option in our
Performer app. I do have background in OpenGL and Inventor, both of
which have you specify whether or not to use double buffering.
Back to the calculation... In a sample configuration, if I'm using
small pixels (256 bits) then after all the multi-samplings, RGBA, and
Z-buffers have been accounted for, I have only used 224 bits.
2 1280 x 1024 Channels on 1 RM-6
4 samples of each pixel
10 bits/component for RGBA
15 bit Z-buffer
1 bit stencil buffer
4 * ( 4*10 + 15 + 1 ) = 224 bits used
256 - 224 = 32 bits/pixels unused
Out of our 80MB of RM memory, we are utilizing 70MB, leaving us with
10MB of unused memory that I suppose could be used as the back color
buffer. Is this the case?
What if a configuration uses ALL of the available RM memory, or just
slightly less than all of it, but not enough to fit an extra "back"
buffer in? Does the system use host memory at that point with a
possible performance hit due to accessing memory outside the RM?
For instance, with 1 1280x1024, 8 samples, 10 bits/component for RGBA,
23-bit Z-buffer, and 1-bit stencil, we use 512 bits/pixel.
8 * (4*10 + 23 + 1) = 512
Of course, this exhausts all 80MB in the RM.
The only remaining possibility, if performer doesn't use double
buffering, would be for it to basically follow the scan converter
filling in the new frame information immediately after it. However,
this seems extremely difficult.
Thanks for any information you may be able to provide.
Pete Willemsen
--
Pete Willemsen
pwillems++at++ford.com
Ford Research Lab
=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
Submissions: info-performer++at++sgi.com
Admin. requests: info-performer-request++at++sgi.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:55:35 PDT